BOOK_ACQUISITION_MINIMIZED.md

pipelines/books/BOOK_ACQUISITION_MINIMIZED.md

Book Acquisition — Minimized (post-build)

Living minimized ledger. Supersedes the raw MUST-ACQUIRE counts in docs/BOOK_ACQUISITION.md

range-by-range, as each range's minimization audit lands. Driver of the cut: **a huge amount of

canon is now built + grounded** (RCM [ACTIVE v1.1.4] covers Ch 2-77 + Pro + Epi; Ch 2-15 spine

entries are all built + gate-green), so most MUST-ACQUIRE information is already synthesized,

calculable, or deep-researchable to high fidelity. Per Josh's binding directive (2026-07-14):

**do not buy a book for information we already have, can calculate, or can deep-research /

infer with high confidence.** Lean hard toward DON'T-BUY; keep BUY only for the genuinely

irreducible §17.1 primary-practitioner floor.

1. Headline

Full-arc result (2026-07-14): all seven ranges are AUDITED against the built spine + the RCM

[ACTIVE v1.1.4] + Josh's owned library (docs/BOOKS_OWNED.md). The ~158-title / ~$6,000 program

collapses to 7 genuine physical BUYs (~$266) — a 96% cut, ~$5,738 saved. Four are non-negotiable

zero-owned-mitigation §17.1 anchors (Fernandez *Bwiti*, Schipper *Taoist Body*, Lehtola *The Sámi People*,

Brown *Mama Lola*); three are care-calls Josh can reduce (Abimbola *Ifá* → owned Drewal; Manker drum-corpus

→ digitized museum plates; Watson → recommend KEEP as the sole Aboriginal-authored voice for a MAXIMUM-care

node). See §8 for the full-arc BUY table, the music/significance finding, and all 34 care-downgrades

awaiting Josh's confirm.

• Confirmed across all 79 nodes: every spine entry (Ch 2-77 + Pro + Epi) is already built + gate-green from the RCM synthesis + deep-research path with none of the ~158 titles purchased — direct existence-proof that the substrate path sufficed for the whole arc. The 7 retained BUYs are exactly the irreducible §17.1 primary-practitioner floor: a living sacred tradition where the author IS the anchor AND no owned companion covers it.
• Projection retired → measured. The Ch 2-15 projection (~10 buys / ~$5,600 saved) is superseded by the completed full-arc count: 7 buys / ~$5,738 saved. The later-range flagship §17.1 anchors did survive as predicted — Ch 21 Taoism (Schipper), Ch 29 Sámi (Lehtola + Manker), Ch 42 Haiti (Brown), Ch 57 Australia (Watson) are the incremental irreducibles beyond the Ch 2-15 pair.
• Sacred-texts USB + the "free PDF" caveat (2026-07-14). Josh's sacred-texts.com USB (~600 public-domain primaries, catalogued in BOOKS_OWNED.md §7) confirms almost every DON'T-BUY and nearly all 34 care-downgrades by putting the actual primary in hand. Critically, the audit's "FREE-ONLINE" items (Munro-Hay Aksum, Kramer *Sumerian Mythology*, Heidel, the full Avesta, the Egyptian funerary corpus, the Meso/Andean + Norse/Celtic + Japanese + Pacific mythology primaries) are mostly OWNED on the USB or genuinely public-domain — NOT dependent on a maybe-available free download. Josh's concern is right: where a "free PDF" isn't reliably live, the substitute is the owned USB primary + deep-research, never a specific free link. So the acquisition requirement does not hinge on any free PDF, and the BUY count does not rise — it is more solid. The USB does NOT replace the 7 insider-voice BUYs (a public-domain Victorian translation is not a living-tradition initiate).

2. GENUINELY-NEED-PHYSICAL — the BUY list

Only titles where deep research cannot substitute AND no owned companion carries equivalent depth: the

author IS the living-tradition anchor and the work is a §17.1 elevated-care primary-practitioner voice.

Now cross-referenced against Josh's owned library (docs/BOOKS_OWNED.md, 2026-07-14) — which confirms

the DON'T-BUY substitutions in §3 are owned-in-hand, and tightens Ch 2-15 to 1 definite BUY + 1 care-call.

Ch 2-15

BUY subtotal: 2 (~$76), or 1 (~$38) if Josh rules Drewal-owned covers the Yoruba §17.1 floor. Both are

the minimal irreducible §17.1 anchors, not immovable. Fernandez is the one title with zero owned mitigation.

3. DON'T-BUY — by class (31 titles)

Every DON'T-BUY title's *information* is obtainable without the purchase. The corpus pipeline

deep-researches all of these at build time (see §5). Substitution paths below.

Owned-library upgrade (2026-07-14): cross-referencing docs/BOOKS_OWNED.md upgrades several of these

from "deep-researchable" to owned-substitute-in-hand — no longer even reliant on deep research:

Turnbull #2 / Bahuchet #3 / Sarno #7 → owned **Hewlett *Congo Basin*; van Beek #5 → owned Griaule

*Ogotemmêli* + *Pale Fox*; Bascom #4 → owned Drewal *Yoruba Ritual*** (+ retained Abimbola); the Ch-6

Coningham/Anuradhapura → owned *Prehistory of Sri Lanka*; Ganapathy (Ch 7) → owned **Zvelebil *Poets

of the Powers*; the Ch-8 Swahili buys → owned Middleton *World of Swahili* + Pouwels *Horn and

Crescent*; Mary Leakey (Ch 10) → owned R. Leakey *Origin of Humankind***; the Ch-10/13 San rock-art →

owned **Vinnicombe *People of the Eland* + Barnard; the Ch-15 Coffin Texts/afterlife → owned Hornung

*Books of the Afterlife* + Aldred *Akhenaten*** + the sacred-texts USB. This makes the §4 care-downgrades

lower-risk (an owned primary now backs the San, Congo, Dogon and Yoruba floors).

ALREADY-HAVE (1)

CALCULABLE (0)

None in this range as a pure class, though two DEEP-RESEARCHABLE titles are effectively published-data

(Stocks' measured Serapeum copper-saw cutting rates; Srinivasan's measured ~1.5% C wootz crucible

metallurgy) — the load-bearing numbers are in peer-reviewed papers, so no purchase.

DEEP-RESEARCHABLE (30) — high-confidence synthesis substitutes the specific text

Non-§17.1 (21):

§17.1 high-care (9) — DON'T-BUY recommended, but each is FLAGGED for Josh's explicit care-confirm (see §4):

4. CARE-DOWNGRADES — Josh must confirm each (9)

CVD §17.1 mandates the primary-practitioner anchor for high-care / elevated-care cultural sources. Moving

any §17.1 high-care title off BUY is a care-risk decision the audit will NOT apply silently. Each below

is *recommended* DON'T-BUY on the substitution paths in §3, but needs Josh's explicit yes before it

drops off the acquire list. The two §17.1 titles kept as BUY (Fernandez *Bwiti*, Abimbola *Ifá*) are the

retained primary-practitioner floor and are not in this list.

Note on the pairings: the Bascom (#4) and Verger (#8) downgrades lean on the retained Abimbola *Bwiti/Ifá*

BUYs carrying the irreducible Yoruba insider voice; the Turnbull (#2), Bahuchet (#3) and Sarno (#7)

downgrades lean on the owned Hewlett *Hunter-Gatherers of the Congo Basin* volume. If Josh wants a

firmer floor, the highest-value single re-instatement to BUY would be one San primary (Lewis-Williams) and

one Congo-Basin primary — but the owned/on-order companions make even those defensible as deep-research.

5. The pipeline deep-researches everything not on the BUY list

Per docs/BOOK_ACQUISITION.md §"Roadmap addition" and docs/SPINE_CASCADE_READINESS.md: the corpus

pipeline treats deep web research + agent deep-dives + adversarial-review synthesis as the default

substrate path for every DON'T-BUY title. A chapter's substrate step pulls its RCM section, acquires only

its (now-minimal) BUY anchors, and deep-researches the rest. The Ch 2-15 build is the existence proof: all

14 spine entries are already built + gate-green with zero of the 31 DON'T-BUY titles purchased.

6. Data-hygiene items surfaced (for the RCR cleanup pass)

• **Hewlett *Intimate Fathers* — stale registry conflict.** RCM T1_RCM [ACTIVE v1.1.4] line 2492 = owned physical; registry row T0_RC_00056 = needed_t1. Reconcile the RCR row to owned_physical.
• **Copenhaver *Hermetica* is a Ch-15 needed_t1 RCR row but its MUST-ACQUIRE slot is assigned to Ch 27** in BOOK_ACQUISITION.md — out of scope for the Ch 2-15 minimization; audit it under the Ch 26-38 range, not here.
• Range count drift: BOOK_ACQUISITION.md labels Ch 2-15 "(34)" but names 33 discrete titles — one bundled/duplicate. Cosmetic; reconcile when the range list is next edited.

7. Status of the full ~158

All seven rows now read DONE. The final full-list result is **locked: 7 physical BUYs, ~$266, ~$5,738 saved

(96%)** — floor of 5 if Josh accepts the two lowest-risk care-calls (Abimbola → owned Drewal; Manker →

digitized drum plates). The full-arc BUY table + music/significance finding + the 34 care-downgrades are in §8.

8. FULL-ARC BUY LIST (Ch 16-77 + Pro/Epi + music + cross-cutting)

SIGNED OFF + PURCHASED (Josh, 2026-07-15): the 7-book BUY list is bought and the 34 care-downgrades are ratified as listed. This section is now the executed record; the acquisition track moves to intake/processing (the book-processing inbox pipeline).

The complete minimized acquisition list for the whole game, all seven ranges audited against the built spine

(all 79 nodes exist + gate-green), the RCM [ACTIVE v1.1.4], docs/BOOK_ACQUISITION.md, and Josh's owned

library (docs/BOOKS_OWNED.md). Every title not on this table is DEEP-RESEARCHABLE, OWNED-SUBSTITUTE, or

CALCULABLE/FREE-ONLINE — the substrate path that already produced all 79 spine entries with zero of the

~158 titles purchased.

8.1 The master BUY table — 7 physical purchases across all 79 nodes

BUY subtotal: 7 physical purchases (~$266 @ ~$38). Floor of 5 if Josh accepts the two lowest-risk

care-call reductions (#2 Abimbola → owned Drewal + Bascom; #5 Manker → digitized museum drum plates). #7

Watson is a care-call but the recommendation is KEEP. #1 Fernandez, #3 Schipper, #4 Lehtola, #6 Brown are

non-negotiable — zero owned mitigation, author-is-the-anchor, living sacred tradition.

8.2 Ranges that yielded ZERO buys — Ch 66-77 + Pro/Epi (12) and cross-cutting/music (1)

Every endgame title is classical / mythological / philosophical scholarship on an **extinct or exhaustively-

documented** tradition — none is a §17.1 living-oral-practitioner floor. All are deep-researchable, public-

domain-free, or owned-adjacent.

Ch 66-77 + Pro/Epi = 0 BUY. Cross-cutting = 0 BUY. (Endgame is mostly game-canon; the Tolkien block,

Plato primaries, and the Eddas/Mabinogion are Frame-A pattern reference or public-domain — free, not

purchase, per BOOK_ACQUISITION.)

8.3 Music / significance finding — 0 incremental buys

• The MUST-ACQUIRE roster contains zero dedicated music / ethnomusicology titles. Nothing across the ~158

is a music book. Musical and sonic substrate rides inside the §17.1 cultural anchors already counted —

Sámi joik inside Lehtola + Manker (Ch 29); Aboriginal songline/sung-law inside the Ch 57 cluster (Watson +

the deep-researched Stanner/Rose); Andean, Amazonian, and Vodou drumming inside their chapter substrate. No

standalone music purchase is irreducible.

• Music is a production asset, not a book buy. Score, diegetic instrumentation, and culturally-grounded

sound design are composed / licensed / field-recorded at build time (a production line, not an acquisition

line) — and the cultural *authenticity floor* for that work is the same §17.1 anchors above, so it adds $0.

• The significance-lens framework (33 lenses, docs/proposals/SIGNIFICANCE_LENSES.md) is a **design

overlay**, not an acquisition line — it selects and weights substrate the pipeline already deep-researches;

0 buys. Net: music + significance = $0.

8.4 Full-arc care-downgrades needing Josh's explicit confirm (34)

Every §17.1 high-care / elevated-care title recommended DON'T-BUY, aggregated across all ranges. None is

dropped silently — each needs Josh's yes before it leaves the acquire list. The 7 retained BUYs are not

in this list (they are the kept floor).

Ch 2-15 (9) — see §4 for the full table: Lewis-Williams *Believing and Seeing* (San); Turnbull *Forest

People* (Mbuti); Bahuchet *Pygmées Aka* (Aka); Bascom *Ifa Divination* (Yoruba); van Beek *Dogon Restudied*

(Dogon); Lewis-Williams & Pearce *San Spirituality* (San); Sarno *Song from the Forest* (Baka); Verger *Notes

sur le culte des Orisa et Vodun* (Yoruba/Vodun); Krog *Country of My Skull* (TRC). *(All owned-substitute-

backed: Hewlett *Congo Basin*, Drewal *Yoruba Ritual*, Griaule primaries, Vinnicombe *People of the Eland*.)*

Care-downgrade total: 34. The clustered high-value re-instatements, if Josh wants firmer floors: one

Zoroastrian primary (Boyce *Zoroastrians*, the accessible single-vol) for Ch 19; Davis *Passage of Darkness*

or Desmangles *Faces of the Gods* for the Ch 42 Maximum node beyond Brown; Luna *Vegetalismo* for the Ch 50

Amazon floor. Rows 23-25 (Popol Vuh ×2 + Florentine Codex) are flagged for completeness but are FREE-ONLINE

practitioner-collaborated translations — no actual care loss, so they are the safest of the 34. Row 28

(Hopi) is not a purchasable monograph — a standing living-culture consultation commitment, not a buy.

8.5 Final full-arc totals

• Total physical BUY across all 79 nodes: 7 — Fernandez, Abimbola, Schipper, Lehtola, Manker, Brown, Watson.
• Estimated cost: ~$266 (7 × ~$38; Manker's 2-vol antiquarian set may run higher, but it is a care-call reducible to free digitized plates).
• Estimated saved vs the ~$6,000 / ~158-title program: ~$5,738 (96%).
• Floor: 5 BUYs (~$190) if Josh accepts the two lowest-risk care-calls (Abimbola → owned Drewal + Bascom; Manker → digitized museum drum plates); Watson recommended KEEP.
• Existence proof: all 79 spine entries are built + gate-green with none of the ~158 purchased — the substrate/deep-research path is validated end-to-end.

Generated by harness/site/structure_site.py — the URL path is the repo path. review root